An iterative dimension-by-dimension method for structural interval response prediction with multidimensional uncertain variables

摘要

Considering that uncertain factors widely exist in engineering practice, the interval model is used to model the uncertain parameters in engineering structures. To solve the structural interval response prediction with multidimensional uncertain variables, an iterative dimension-by-dimension method (IDDM) is proposed. IDDM seeks the minimum and maximum points of the uncertain variables dimension-by-dimension and updates the nominal value of other uncertain variables through an iterative process. Thus, IDDM is efficient for structural interval uncertainty analysis, especially for problems with multidimensional uncertain variables. Both the vertex method and Chebyshev interval method are improved through IDDM. In addition, the serial strategy and parallel strategy are developed to conduct the IDDM; the serial strategy needs fewer iterative steps to reach convergence by using the newest information in the iterative process: while the parallel strategy calculates much faster for just one iteration process through the parallel processing. Both numerical and engineering examples are used to verify the validity of the proposed method. (C) 2019 Elsevier Masson SAS. All rights reserved.